[JENKINS-29328] NullPointerException parsing a Background section that has a Before hook.

[stopiccot]

After recent update of our testing setup to cucumber-ruby 2.0 jenkins json parsing step started to fail. It turned out to be a known issue with a JIRA task for it. Here is simple fix for that.

[jhansche]

@jtnord any objections to merging this?

[jtnord]

@jtnord any objections to merging this?

Yes - it changes an existing test (by changing the json) that still needs to pass - also not convinced that the UI will be updated accordingly to show this

[stopiccot]

Yeah, I've changed existing json to match json's generated by cucumber-ruby 2.0+, got test failure and then updated code so test passed again. You think we need separate test case for this?

[JeanMertz]

At our company, we just ran into this issue as well. Hoping to see this PR move forward soon.

[bbossola]

It's really time to have this merged. In the meantime I just rebuild the hpi myself, but it's unfortunate that you are not officially releasing this

[edwardvfluke]

Any update on this merge into master?

[benSlaughter]

@jtnord What are your requirements for this fix?
I have been following for some time, and I have this issue.

Are some extra tests required to cover other use cases?
@bbossola Have you been running the custom HPI with success?

[jtnord]

@benSlaughter

it changes an existing test (by changing the json) that still needs to pass - also not convinced that the UI will be updated accordingly to show this

There also needs to be a test to show that a failire in a before hook in a background section will correctly report the failing test and ideally render it in the UI (which at the moment I think is unaddressed by this change).

I would also really like the cucumber guys to define the gherkin json output spec and stick to it rather than changing it on a whim and making each tool do different things. The output of the different tools (cucumber-jvm, cucumber-ruby etc is all different....)

[kingdonb]

I do have a lot of Background steps, as well as Before hooks and AfterStep hooks, and I do want to know which tests have failed because of problems executing the Background steps. So I agree that this should not be merged as-is.

Thanks for this PR, I've built the plugin from it and subscribed to this issue.

If there is another release without merging a fix for this, would be good to note prominently on the release page that using Cucumber 2.0.1+ with Background steps and before hooks can trigger this.

(Note: this PR did manage to prevent the crash bug for me. 👍 )

[intrigeri]

What are the remaining blockers (if any)?

[kingdonb]

Still needs someone to Add a test that shows what happens when a Background step caused a test to fail before the Scenario steps have begun.
(I am fairly sure it will not be reflected correctly in the detail. I don't think it will cause the whole report to fail, I'm not sure if it will accurately reflect a failure)

I can probably add this test myself, but I'm not familiar enough with the Java to get this fixed to work optimally.

[damontgomery]

Are there any updates around this? We're hitting similar issues with recent versions of Cucumber and the plugin.

[kingdonb]

@damontgomery The change was never merged back into a release, because it wasn't really complete. The PR fixes the crash bug though. It might be best to just merge the thing and create a new release from it, even if it's not complete, it fixes a crash bug and it's a bug that probably affects 80% or more who tried using this plugin.

Personally I use the other Cucumber plugin for Jenkins because it didn't have a crash bug. The fix has some serious issues still, but they are not as bad as a crash bug.

(You can build the plugin from the PR branch from @stopiccot if you need to get it installed and get to work right now)